Work plan

  1. write up some stuff I need to know about MDV surface moisture and its behaviour in this RMD as input to the paper

  2. explore soil moisture dataset

    • data cleaning necessary?
    • understand surface moisture and temperature information distributions, determine cut-off value (0°C? -x°C?)
    • spatial configuration of loggers, i.e. think about validation strategy (how to do the spatial CV)

  3. data gathering for pre-study: find a cloud free day and get as many useful spatial predictor datasets as possible

  4. describe and understand relations between surface moisture and predictors & temperature

    • surface moisture and elevation
    • surface moisture and temperature

  5. build model for case study

  6. run workflow for the whole temporal and spatial setting

To do’s:

  • ask Pierre about salinity and pH-paper progress

Paper relevant info

Introduction

Research Question:

  • For Method 1 and 2: How accuarate can surface moisture be modeled for the open soil areas within the Antarctic Dry Valleys?

  • actually interesting question: Which factors does the spatio-temporal surface moisture configuration in the MDV depend upon?

What we know about surface moisture in the MDV

Take a look at Overleaf note collection

Questions

  • should I only use values above 0 degrees? Find out when water freezes in the MDV and how the relation to salinity would be

  • How does RH relate to soil moisture?

  • can we detect Soil Moisture directly with RS techniques? or is indirect modelling needed?

Data

Calibration and validation

AWS and iButton spatial and temporal distribution

Already available:

  • iButton data

To download:

Antarctica Soil Climate Research Stations

Data can be found here.

Get following Variables:

ID, Year, Hour, Date Air Temp (°C, 1.6m) Solar Radiation (W/m²), 3m Wind Speed (m/s), 3m Wind Direction (azimuth), 3m

Soil Temp (°C) Surface under white rock Soil Temp (°C) Surface under black rock

Soil RH (%) 5cm Soil RH (%) 15cm

MRC Soil Temp °C 0 in MRC Soil Temp °C 3 in 7,6cm

Soil 2cm: ER unitless Real dielectric constant EI unitless Imaginary dielectric constant Temp °C Temperature ER_COR unitless Temperature corrected ER EI_COR unitless Temperature corrected EI WATER wfv Soil moisture SALINITY g NaCl/L Soil salinity SOIL_COND S/m Soil conductivity SOIL_COND_COR S/m Temp corrected soil conductivity WATER_CON_COR S/m Temp corrected conductivity of the water in the soil

Example: Explorer’s Cove

Could be interesting to get precipitation from there

Potential spatial predictors

Potential spatial predictors

  • backscatter from a SAR satellite (Sentinel 1? good temporal and spatial resolution,ongoing mission since April 2014, open access) as a direct measurement of SMC
    • Radarsat-2 for iButtons
    • Sentinel 1a and 1b for climate stations 2014-2019
    • Radarsat-1 for climate stations before 2011?
  • LST to capture SMC effect on thermal inertia since 1999 30m resolution
  • wind speed and direction as forcing for thermal behaviour of wetted soils. available in NZ: 3,3 km resolution from 2013-2020 and 0,89 km resolution from 2017-2020
  • optical satellite data (mostly longer wavelenghts interesting, i.e.SWIR to capture SMC via darkening of surfaces) Landsat / Sentinel 2 (since 2015), try approach as in Haubrock et al. (2008): R2 values gained from linear correlation between soilmoisture and normalized difference of two bands
  • terrain-derived data
    • TWI for (subsurface) routing of meltwater (calculated on 8m)
    • capture regional microclimate i.e. small depressions, polyon cracks etc. (calculated on 1m)
  • distance to lakes and runoff channels to capture wetted margins
  • soil type and permafrost distribution data
  • pH (Pierre: “pH is very much correlated to soil moisture, and more linear than EC(because it’s already a log scale!), but I could also share EC estimates if need be.”) electrical conductivity as salinity proxy which may interfere with backscatter

Radar data

For Radar only Radarsat satellites and Senintel 1 (C-Band) would be useful: https://www.unavco.org/instrumentation/geophysical/imaging/sar-satellites/sar-satellites.html

For SAR processing via python: https://pyrosar.readthedocs.io/en/latest/

download SAR data from https://search.asf.alaska.edu (Radarsat 1 is on there)

Radarsat-2 ESA archive - 2 scenes from the ice tongue… is that all there is? Looks better: overview Radarsat 2: Radarsat 2 Sentinel 1: Sentinel 1 https://earth.esa.int/eogateway/catalog/radarsat-1-2-full-archive-and-tasking Eo Sign in https://earth.esa.int/web/guest/pi-community/myearthnet Data access request filed, view progress under https://esatellus.service-now.com/csp?id=esa_homepage

Is Radarsat and Sentinel 1 comparable? Then I could use Radarsat 1 and 2 and Sentinel 1 Both C band and multiple polarizations

Pre Study: 2011-01-01 to 2014-01-01

Already available:

  • DEM 8 / 30m * TWI * slope
  • rock outcrop to use as a mask
    • use soil type map to find out where there is only rock and no soil
  • soil types
  • pH model (Pierre: “pH is very much correlated to soil moisture, and more linear than EC (because it’s already a log scale!), but I could also share EC estimates if need be.”)
##  [1] "DEM_8m_MDV_clean_aoi_filled_filtered_new.tif"
##  [2] "DEM_8m_MDV_filled_aoi.mgrd"                  
##  [3] "DEM_8m_MDV_filled_aoi.prj"                   
##  [4] "DEM_8m_MDV_filled_aoi.sdat"                  
##  [5] "DEM_8m_MDV_filled_aoi.sdat.aux.xml"          
##  [6] "DEM_8m_MDV_filled_aoi.sgrd"                  
##  [7] "DEM_8m_MDV_filled_aoi.tif"                   
##  [8] "full_size_grids_all_layers.RDS"              
##  [9] "predcrs.rds"                                 
## [10] "radar"                                       
## [11] "Wind_Expo.mgrd"                              
## [12] "Wind_Expo.prj"                               
## [13] "Wind_Expo.sdat"                              
## [14] "Wind_Expo.sdat.aux.xml"                      
## [15] "Wind_Expo.sgrd"                              
## [16] "Wind_Expo_900m.mgrd"                         
## [17] "Wind_Expo_900m.prj"                          
## [18] "Wind_Expo_900m.sdat"                         
## [19] "Wind_Expo_900m.sdat.aux.xml"                 
## [20] "Wind_Expo_900m.sgrd"

To acquire / produce

  • RS data:
    • SWIR
      • Landsat 8 launched Feb 2013,
      • Landsat 7 bands 5 and 7 similar to SWIR in L8, since 1999, sensor noise issue,
      • Landsat 4-5 bands as in 7 since 1984-2013
    • downscaled LST from 1999 on
    • Radarsat-2 data
  • EC (Pierre)?

Methods

  • Statistical Exploration of associations
    • check wind relevance with station data?
  • Modelling
  • Explainable AI to find out which factors are relevant how
  • Analysis of Moisture patterns

Method schematic overview

Discussion

References